With the increasing of depth and complexity of the convolutional neural network, parameter dimensionality and volume of computing have greatly restricted its applications. Based on the SqueezeNet network structure, this study introduces a block convolution and uses channel shuffle between blocks to alleviate the information jam. The method is aimed at reducing the dimensionality of parameters of in an original network structure and improving the efficiency of network operation. The verification performance of the ORL dataset shows that the classification accuracy and convergence efficiency are not reduced or even slightly improved when the network parameters are reduced, which supports the validity of block convolution in structure lightweight. Moreover, using a classic CIFAR-10 dataset, this network decreases parameter dimensionality while accelerating computational processing, with excellent convergence stability and efficiency when the network accuracy is only reduced by 1.3%.

In a wireless sensor network, the sensor nodes transmit the acquired information to the server through the data transmission link. On the serverside, the data are processed, fused, and expressed to serve the user. Sensor deployment is a key factor related to the stability and security of wireless networks. This article uses environmental changes to drive related technologies to deploy wireless sensors. In this article, environmental change-driven means that through certain deployment cost model assumptions and problem descriptions, network deployment is artificially divided into two stages: initial deployment and redeployment. In the deployment phase, by referring to the idea of virtual force, a new sensor deployment algorithm is proposed in the redeployment phase, which can well solve the stability- and security-related issues encountered in agricultural wireless sensor networks. In this algorithm, the moving distance of the mobile receiver and the average coverage of the network are calculated based on the virtual force, the direction, and the number of adjacent clusters. Finally, the algorithm model was simulated in MATLAB, and the feasibility of the algorithm was verified by analyzing the event coverage and the moving distance of nodes. The final simulation results show that the algorithm proposed in this paper can achieve better performance than existing algorithms in terms of average coverage and moving distance.

This article focuses on the demand for sensor access in the sensing layer of the agricultural IOT. 12 soil moisture sensors from China, the United States, Germany, the United Kingdom, and Australia were selected, with the aim of directly connecting the sensors to the IOT. A comparative study was conducted on their communication interfaces and protocols. The 12 selected soil moisture sensors can be divided into three categories based on interface types: analog interface, SDI-12 interface, and RS485 interface. We conducted signal acquisition and communication protocol research on these three types of interfaces, and using the STM32 microcontroller as the platform, the data collection and conversion of these three interface sensors were programmed according to their respective communication protocols, and corresponding soil moisture data was obtained. Research has found that Modbus protocol is the simplest and most straightforward. SDI-12 protocol is second in complexity. Analog interfaces, rely on the calibration formula provided by the manufacturer or require self calibration for their measurement data accuracy.